Products useful for determining or monitoring the condition or remaining useful life of productive assets, including but not limited to batteries used in military aircraft, most often perform this surveillance function by evaluating signal or data values obtained during asset operation. One means for determining or monitoring the condition of an asset involves estimating the expected data values and comparing the estimated values to current or present data values obtained from the asset. When the estimated data values characterize the desired or expected operation of the asset, a disagreement between the estimated data values and the present data values provides a sensitive and reliable indication of an asset degradation or fault condition and can further provide an indication of the particular cause and severity of the asset degradation or fault. The disagreement between each estimated data value and each present data value can be computed as the numerical difference between them. This difference is often referred to as a residual data value. The residual data values, the present data values, or the estimated data values can be used to determine condition of the asset and to identify or diagnose asset degradation or fault conditions.
In recent years, the field of prognostics has reached buzzword status. The result of which has been an avalanche of literature describing many different prognostic algorithms that are supposedly capable of estimating the remaining useful life (RUL) of an individual asset. However, upon closer examination, it is evident that the current state of the art is cluttered with methods that either do not produce estimates of the RUL or do not provide a realistic method for relating degradation to the RUL for an in-service asset, such as military aircraft batteries.
For example, a general path model (GPM) (Lu, C. Joseph and William Q. Meeker, “Using Degradation Measures to Estimate a Time-to-Failure Distribution”, Technometrics, Vol. 35, No. 2, pp. 161-174: May 1993.) is founded on the concept that a degradation signal collected from an individual asset will follow a general path until it reaches an associated failure threshold. Since its introduction, the thought model proposed in the GPM has been prolifically adopted by modern researchers and has resulted in a plethora of techniques that can be related to the GPM in one way or another.
There are two fundamental assumptions of the GPM and its modern counterparts. First, there exists a path for the degradation signal that can be parameterized via regression, machine learning, et cetera and secondly, there exists a failure threshold for the degradation signal that accurately predicts when an asset will fail. For modern computational capacity, the first assumption is minor, in that many methods exist for parameterizing simple (polynomial regression, power regression, et cetera) and complex (fuzzy inference systems, neural networks, et cetera) relationships from data. The assumption of the existence of a threshold that accurately predicts asset failure is not so easily reconciled. While the existence of a failure threshold has been shown to be valid for well understood degradation processes (for example, seeded crack growth) and controlled testing environments (for example, constant load or uniform cycling), Liao observes that for real world applications, where the failure modes are not always well understood or can be too complex to be quantified by a single threshold, the failure boundary is vague at best (Liao, Haitao, Wenbiao Zhao, and Huairui Guo, “Predicting Remaining Useful Life of an Individual Unit Using Proportional Hazards Model and Logistic Regression Model”, Proceedings of the Reliability and Maintainability Symposium (RAMS), pages 127-132: Jan. 23-26, 2006). Wang, et al. attempt to address this problem by integrating uncertainty into the estimate of the threshold (Wang, Peng and David W. Coit, “Reliability and Degradation Modeling with Random or Uncertain Failure Threshold”, Proceedings of the Annual Reliability and Maintainability Symposium, Las Vegas, Nev.: Jan. 28-31, 2007), but in the end the authors replace an estimate of the threshold with another, more conservative estimate.
For the most part modern prognostic methods have failed to actually produce estimates of the RUL; however, it is important to note that there are methods available that actually estimate the RUL of an individual asset. For example, most notably Bonissone, et al. (Bonissone, P. and K. Goebel (2002), “When Will It Break? A Hybrid Soft Computing Model to Predict Time-to-Break Margins in Paper Machines”, Proceedings of SPIE 47th Annual Meeting, International Symposium on Optical Science and Technology, Vol. 4785, pages 53-64: 2002) use a complex system involving many statistical and artificial intelligence based methods to infer the RUL of a paper machine. However, the sheer complexity and poor estimate accuracy limited the applicability of this work to an academic forum.
Hence, there is a need for a system and method for prognosticating the remaining useful life (RUL) of an asset, such as batteries used in military aircraft, that ameliorates or overcomes one or more of the shortcomings of the known prior art.